It was quite a pain - ArrayBuffers / Typed Arrays are fussy about byte alignment - but seems to be worth it.

Parsing times are down (modest gains for small objects / Firefox, bigger gains for large objects / Chrome, performance gains from 10% - 200%).

Also as a positive side effect, changing interleaved face arrays into uniform arrays seems to have helped a bit with gzip compression (about 6% smaller size for gzipped 526K Lucy).

Proper IcosahedronGeometry next...

It makes the whole thing imprecise, but cubes are now properly reflective.

Also manually fixed buffalo model and test scene to be valid JSONs.

Work-in-progress => do not merge yet.

OBJ converter, JSONLoader, BinaryLoader and some examples are done. Still need to do Blender exporter, SceneLoader and more examples.

Advantages:

- faster parse times
- bypass array initializer size limit in Firefox
- no more crashes in Chrome 15 (caused by too many concurrently active workers)

Disadvantages:

- more browser UI freezing
- multiple models tend to appear all at once with the last one (instead of popping up progressively)

And still readable code...

It's enough to clone just for hits, no need to do it for every face.

Removed some clone() calls while still keeping the code somewhat pretty.

So we can set it to lowp for mobile devices if needed.

Not sure what was that supposed to be for, we didn't have a single use case.

Also fixed RTT example, it was affected by unlit Lambert change.

Zombies now look correct.

Though still getting undebuggable random crashes in Chrome with ro.me animals demo :S

Mostly just shuffling code around to have more organized "big picture" view.

https://github.com/mrdoob/three.js/pull/689

More moving ifs out of inner loops.

Also here moved ifs out of the inner loop.

Starts to get hard to measure performance impact - baseline JavaScript stuff just to produce some change gets very costly. Simply calling Math.random() hundreds thousands times causes massive garbage collection :S.

No need for copy of attributes map here, attributes list has the same function.

More moving ifs out of inner loops: gained 25% performance for dynamic meshes with custom attributes.

Hope I didn't break something.

Moved most of ifs out of inner loops for ~25% performance gain with dynamic meshes. Custom attributes setting for meshes can still be improved more.

This gave us ~13% performance gain for dynamic meshes.

Interestingly, this was breaking OpenGL but not ANGLE.

Actually, this would break cameras added as childs of objects, reverting.
So there is no clear and simple way to handling multiple scenes unfortunatelly.
Encouraging `scene.add( camera )` seems like the right thing to do then, less problems.

Actually, I think this can be handled magically instead of requiring understanding of the concept that the camera is part of the scenegraph.
While I was at it, I fixed it so you can use a camera for rendering different scenes (the renderer will add it and remove from each scene graph)	.

Fix was trivial, I guess for the moment they can stay.